Application of machine learning approach for iron deficiency anaemia detection in children using conjunctiva images

Abstract

Iron deficiency is commonly referred to as anaemia which is a general public health problem that normally occurs as a result of a reduction in red blood cells which is common in developing countries such as Africa. In this study, machine learning algorithms such as CNN, k-NN, Naïve Bayes, Decision Tree and SVM were utilized for the study to detect anaemia in children using conjunctiva images. The images were segmented into their various CIELAB colour space components and the ROI from each image was retrieved. The dataset was split randomly into 70:10:20, which were then used to train, validate, and test the models, as appropriate. The CNN achieved the highest accuracy (98.45 %). The findings of this study demonstrate that non-invasive techniques are essential for detecting anaemia in children. This study deploys a cost-effective mechanism, and result-orientated, to detect anaemia in developing communities where health facilities, resources, and personnel are scarce.